Add FoxNoseLoader and create_foxnose_tool

[loookashow]

Summary

• Add FoxNoseLoader — a LangChain BaseLoader for bulk document loading from FoxNose folders with cursor-based pagination, lazy loading, and async support
• Add create_foxnose_tool() — a factory function that wraps FoxNoseRetriever as a LangChain tool for use with LLM agents
• Update README with new component descriptions, usage examples, and documentation links

Details

FoxNoseLoader

• Iterates over all resources in a folder using cursor-based pagination
• Supports load(), lazy_load(), and alazy_load() for memory-efficient streaming
• Configurable batch size, filtering, sorting, content mapping, and metadata control
• Factory method from_client_params() for convenient initialization

create_foxnose_tool

• Creates a LangChain tool from a new or existing FoxNoseRetriever
• Customizable tool name and description for agent integration
• Two response formats: plain string or (str, list[Document]) tuple
• Compatible with any LangChain agent framework (e.g. langgraph)

Test plan

• Unit tests for FoxNoseLoader validation and functional behavior
• Unit tests for create_foxnose_tool validation and functional behavior
• Shared fixtures extracted to conftest.py
• Existing retriever tests still pass